Cloudflare announced that from now on, programming AI agents can complete Cloudflare account opening, paid subscription activation, domain name registration, and obtaining API tokens for deploying code on behalf of users. The entire process does not require manual entry into the console or manual filling in of credit card information; in this new model, humans usually only need to intervene once during the confirmation of the terms of service.
In the past, in order for an AI agent to actually deploy a project to a production environment, it usually required three prerequisites: available accounts, payment methods, and API tokens. These key steps have long had to be handled manually. Now, Cloudflare is gradually handing over these capabilities to agents for execution, allowing AI to not only write code, but also complete the entire set of basic preparations before going online.
This process is based on a protocol jointly built by Cloudflare and Stripe and is part of the Stripe Projects release. Its operating mechanism mainly consists of three links: the first is "discovery", where the agent queries the service directory and confirms the callable services; the second is "authorization", where Stripe is responsible for verifying the user's identity, and Cloudflare will automatically associate existing accounts, or create new accounts when necessary; and the last is "payment", where Stripe provides tokenized payment vouchers instead of real credit card numbers, so the agent can complete the purchase operation without being exposed to sensitive bank card information.
Under default settings, AI agents are capped at $100 per month in spend per service provider. If users want to increase the limit, they can manually increase the limit in their Cloudflare account, or set a budget reminder for higher expenditures to control usage risks and costs.
According to Cloudflare's description, the entire process only takes a few minutes to complete. Users first need to install the Stripe CLI with the Stripe Projects plug-in, then log in to Stripe, run "stripe projects init", and then issue the task of building and deploying the application to their AI agent. Afterwards, the agent will automatically query the service catalog, identify Cloudflare related services, create accounts on demand, purchase domain names, deploy code, and finally return an available application that is already running on the real domain name.
In addition to the automation capability itself, Cloudflare also announced an incentive for startups: any new startup that is registered through Stripe Atlas will have the opportunity to receive Cloudflare points support totaling up to $100,000. This means that Cloudflare hopes to use this new mechanism to further attract early development teams and AI-native entrepreneurs into its platform ecosystem.
It is worth noting that this agreement is not limited to the Stripe system. Cloudflare said that any platform with a logged-in user system could theoretically play a role similar to Stripe and integrate with Cloudflare in the same way. The company also revealed that it hopes to attract more partners to join, and is currently working with Planetscale, with the goal of allowing users to configure Postgres databases directly through Cloudflare in the future.
From the perspective of industry trends, the capabilities launched by Cloudflare this time mean that the AI agent is moving further from an "auxiliary programming tool" to an "operational entity that can execute the online process." When account creation, payment authorization, resource configuration and application deployment can all be completed by agents with limited manual approval, the automation of the software delivery process undoubtedly takes a step forward.